[aaimnr]

Giulio Tononi is the guy who's arguably brought the most interesting perspective on consciousness (information integration theory) since the original Chalmers' problem statement.

Here's him explaining why the problem is hard and how it could be approached, in the middle of some kind of artifical jungle: https://youtu.be/Vl8J3K_ZLkg?t=5m50s

[sbierwagen]

Note that IIT was heavily criticized by Scott Aaronson for producing largely nonsensical results when applied to a simple square grid of XOR gates: https://www.scottaaronson.com/blog/?p=1823 https://www.scottaaronson.com/blog/?p=1799

It is clear to me that, whatever it is we're talking about when we're talking about consciousness, an expander graph doesn't have it.

[visarga]

IIT is missing the self-replication requirement. If a system is a self-replicator, it needs resources and has to avoid dangers. This in turn creates a necessity for perception and ability to select good actions depending on situation. A square grid of XOR gates has none of that.

[goldenkey]

What you mentioned is part of evolution's natural selection through selective pressures. But it isn't precluded that an AI could be created with none of these biological hallmarks of past evolution.

[KingMob]

Former consciousness neuroscientist here. There's some great explanatory abilities about IIT and Tononi's phi measure, but it's not clear it's sufficient.

On the upside, it explains why the cerebellum, despite comprising half the neurons of the brain, has virtually no impact on awareness when removed (like for tumors or epilepsy). The IIT answer is that the cerebellum is highly regular, like a GPU having many units, but all doing the same thing. In this sense, it has lower phi than the cerebrum, which is way more heterogeneously organized. This might also explain why awareness is lost in deep sleep or epileptic seizures; the theory is that the electrical pattern becomes much simpler, and lower phi.

The downside is that it's not clear where the dividing line between conscious/unconscious should be. A planarian only has ~8k neurons; is its phi sufficient for consciousness, or is it a biological robot? Or put it the other way: the phi of things like the internet or a biosphere could be quite high, but are they conscious?

As my advisor liked to joke, "What's the phi of the population of China?"

[visarga]

> "What's the phi of the population of China?"

Small, because if you cut it in 100, you still get 100 functioning parts. Can't cut the brain in 100 and still get functioning mini-brains.

[KingMob]

Ah, but phi measures integration levels, not independent survivability. The cause-effect structures are reduced by 99% in your scenario. The joke (and implied criticism) is that society itself (not the individual members) might have a high enough phi to pass some "consciousness threshold", and if we think that's absurd, it should cause us to question IIT.

Don't get me wrong, IIT is one of the best mathematical models of consciousness out there, but I don't think it's the final word in the matter.

[21]

> The IIT answer is that the cerebellum is highly regular, like a GPU having many units, but all doing the same thing.

Isn't the cortex also the same unit (the cortical column) repeated over and over again?

[KingMob]

Not at all. I wouldn't even argue that the cortical column is the main organizational unit.

[visarga]

I used to consider Tononi as the best philosopher of consciousness until I learned more about neural nets and watched the RL course [1] by David Silver (co-author of AlphaGo).

After I understood the RL paradigm, I realise that Tononi's explanation barely scratches the surface. Yes, there is integrated information, but how does it come about? What is its purpose?

The answer is simple - painfully simple - the goal is to maximise rewards. One goal we all have is to live and have children - and this root goal (a necessity of the genes to propagate, actually) is what guides the evolution of integrated information in the brain. But the environment plays a crucial part in the contents, structure and complexity of consciousness. Integrated information is very dependant on the environment. Yet Tononi & co. still search for it in the brain, as if you can speak of a brain without considering its experiences, and consider experiences without thinking about the world and the problems the agent has to solve.

Just watching reinforcement learning agents learn and evolve in simulated environments, as we had the opportunity for the last 3-4 years, is enough to create a perspective about agents that is not human centric and that is very useful in thinking more clearly about consciousness. You can see a humanoid learn a gait that is like the Ministry of Silly Walks [2], you can see bots playing FPS games, AlphaGo playing against itself, cars driving themselves... That puts human learning and human agenthood in perspective.

[1] https://www.youtube.com/playlist?list=PL7-jPKtc4r78-wCZcQn5I...

[2] https://youtu.be/g59nSURxYgk?t=88

[ozy]

Reinforcement based learning requires self-observation. Especially when done with predictive modeling. Both the brain clearly does. You might like this paper: https://psyarxiv.com/387h9

[visarga]

Good paper. Are there more philosophers that support this view? because it seems to be largely ignored.

[aaimnr]

What does learning have to do with consciousness? These are orthogonal issues. That's the whole point of Chalmers argument.

[visarga]

Well it has a lot to do. We're not born with fully functional minds, we learn our mental skills as we grow. Learning shapes the very concepts we use for representing sensations and thinking. Consciousness is not something 'secreted' in the brain, it's the loop made of 'agent + environment', where the purpose is to maximise rewards. There is no consciousness in itself, just consciousness of something. Learning is what ties together agent and environment, it's the building force of consciousness.

And Chalmers is a dualist that believes there are two realms that can't be explained, and that's ridiculous in this day and age. He's the worst philosopher of consciousness because he led a generation astray with sterile dualist concepts - and where has he led philosophy? Nowhere, there was no insight, no discovery after the "hard problem" because, darn, it's "hard" which is another word for dualism today.

I take Tononi and Dennett over Chalmers any day, but I prefer Reinforcement Learning over all of them as my intuition pump with regard to consciousness. Philosophy is mired in a swamp of bad concepts that are almost useless, they should just use a learning based terminology which is so much more effective. Engineers and experimental scientists create bots that beat humans at Go, a game that can't be brute forced, and they don't realise they've been outrun in their 2000 year marathon by a hundred year old concrete approach. The difference is that RL has the right concepts and philosophy uses extremely refined but ultimately useless concepts. They've realised words don't mean anything in the absolute sense (they all rely on each other, cyclic referential) and are just part of a game, but are still neck deep in useless words instead of using evolutionary and RL concepts to concretely model consciousness and the game it plays.

[aaimnr]

"They've realised words don't mean anything in the absolute sense (they all rely on each other, cyclic referential) and are just part of a game, but are still neck deep in useless words instead of using evolutionary and RL concepts to concretely model consciousness and the game it plays."

That's a full-on nihilistic postmodernism. The fact that words mean something only in reference to other words doesn't have to mean that they are useless. Quine and other pragmatists (Buddhism does the same) argued otherwise - that concepts/theories derive meaning or truth-value from how useful they are in the real world (as a network, rather than individually).

Treating all philosophers as a one camp vs science is mistaken. Whatever any particular scientist or engineer say, there always will be some philosophical assumptions behind it. It's always better to make them explicit rather than be in the dark. The best scientists in history were pretty deep in philosophy as well.

Eg. Tononi is both philosopher and a scientist. He's clearly on Chalmers' side philosophically, he perceives consciousness as something fundamental, MUCH more fundamental than learning. He posits that even stable systems (so with no learning at all) can be conscious. Which makes a lot of sense from phenomenological point of view. He also adds a theory of how specifically consciousness may be related causally with the physical world. That's the scientific part.

Silvers, on the other hand, and the whole RL thing is not concerned with consciousness AT ALL! It's a completely different problem. Actually it may be the case that most of the learning processes in human mind are unconscious!

"There is no consciousness in itself, just consciousness of something. Learning is what ties together agent and environment..." Exactly - if you define learning as a relationship between a system and its environment, you don't need anything else (like cosciousness), just the actual and potential interactions.

Late Wittgenstein, Heidegger, Merleau-Ponty and others would be on the same page with you here, so again, let's not throw the baby of philosophy out with the bathwater. These observations were made in the first half of XX century. They apply perfectly to the naivete of old school symbolic AI (and a logical positivism philosophical stance behind it), as captured by Hubert Dreyfuss, who described all the problems with it from philosophical (phenomenological specifically) standpoint in "What computers can't do" and his more recent paper ( http://cspeech.ucd.ie/Fred/docs/WhyHeideggerianAIFailed.pdf ). RL seems to be step in the right direction in this perspective. However...

"[Learning] ... it's the building force of consciousness."

Well, this part just doesn't make sense. You want to focus on explaining learning? Fine. Do some work on RL, it's enligthening for sure. I completely agree that it's fascinating how new concepts emerge in AlhaGo around some specific baord configurations. It changed people's understanding of the game. But please, don't conflate it with consciousness. And if you do, be open about it and name your position in terms of Chalmers' recent paper. Is it some form of illusionism? Only then we can have meaningful conversation about your actual position on what consciousness is.

Whatever is the relationship of concepts and sensations, however these two aggregates relate to each other and evolve in the mind, consciousness seems to be something more fundamental. Are you saying that AlphaGo is already conscious? If not, can it be made conscious? How? By adding more CPU? A webcam? We can't escape these questions.